fix(sentry): filter out repo metadata when sending events (PROJQUAY-9436) (#4723)

fix(sentry): filter out repo metadata when sending events

test/test_exceptionlog_sentry.py

@@ -386,3 +386,29 @@ class TestSentryBeforeSendFilter:
         }
         result = _sentry_before_send_ignore_known(event, {})
         assert result == event
+
+    def test_filter_401_with_redis_in_repository_name(self):
+        """401 errors should be filtered even if repository name contains 'redis'."""
+        event = {
+            "logentry": {
+                "formatted": "Error 401: Invalid Bearer [token] format; Arguments: {'repository': 'bitnami/redis-exporter', 'status_code': 401}"
+            }
+        }
+        result = _sentry_before_send_ignore_known(event, {})
+        assert result is None
+
+    def test_filter_401_with_database_in_repository_name(self):
+        """401 errors should be filtered even if repository name contains 'database'."""
+        event = {
+            "logentry": {
+                "formatted": "Error 401: Invalid Bearer [token] format; Arguments: {'repository': 'vpenx/clamav-database-mirror'}"
+            }
+        }
+        result = _sentry_before_send_ignore_known(event, {})
+        assert result is None
+
+    def test_keep_actual_redis_error(self):
+        """Actual Redis errors should NOT be filtered."""
+        event = {"logentry": {"formatted": "Redis connection refused: ECONNREFUSED 127.0.0.1:6379"}}
+        result = _sentry_before_send_ignore_known(event, {})
+        assert result == event

util/saas/exceptionlog.py

@@ -79,6 +79,10 @@ HTTP_4XX_PATTERN = re.compile(
     re.IGNORECASE,
 )
 
+# Regex pattern to strip Arguments metadata from error messages
+# Matches: "; Arguments: {'repository': 'bitnami/redis-exporter', ...}"
+ARGUMENTS_PATTERN = re.compile(r";\s*Arguments:\s*\{.*\}$", re.IGNORECASE | re.DOTALL)
+
 BROWSER_ERROR_PATTERNS = [
     "script error",
     "syntax error",
@@ -146,6 +150,9 @@ def _extract_searchable_text(ex_event: Any) -> set[str]:
     if "culprit" in ex_event:
         texts.add(str(ex_event.get("culprit", "")).lower())
 
+    # Strip Arguments metadata to prevent false positives from repository/image names
+    texts = {ARGUMENTS_PATTERN.sub("", t) for t in texts}
+
     # Remove empty strings
     texts.discard("")